Datacurve, a Y Combinator-backed startup targeted on constructing superior datasets for AI and software program growth, has closed a $15 million Collection A spherical led by Chemistry. This recent capital follows an earlier $2.7 million seed increase, bringing the corporate’s complete funding to round $17.7 million.
Based by Serena Ge and Charley Lee, Datacurve goals to resolve a essential bottleneck in AI coaching: acquiring advanced, real-world knowledge that goes past easy coaching units. The corporate’s platform produces research-grade coding challenges, debugging duties, and personal repository benchmarks designed to assist AI fashions enhance reasoning, problem-solving, and coding efficiency.
Datacurve’s distinctive bounty-based contributor system, Shipd, engages prime engineers, together with expertise from DeepMind, OpenAI, Anthropic, and Vercel, to submit high-quality datasets by structured challenges. Up to now, Shipd has distributed over $1 million in bounties, creating an incentive-driven market for helpful knowledge contributions.
“We deal with this as a shopper product, not a knowledge labelling operation,” mentioned Serena Ge, co-founder and CEO. “We spend a number of time optimising the expertise to draw and retain the engineers whose contributions matter most.”
With AI fashions changing into more and more subtle, the necessity for extra nuanced post-training datasets is rising quickly. Datacurve’s knowledge fills this hole by offering analysis and fine-tuning assets important for real-world mannequin efficiency enhancements.
Wanting forward, Ge and Lee plan to scale their group and platform additional, with ambitions to increase past code knowledge into sectors like finance, advertising and marketing, and healthcare.
